Its been a rough period on the investment side of your business hasn’t it? And those pre-1999 claim debts have to be paid sometime. So maybe levies need to rise across the board. But why are your proposed funding adjustment increases not uniform across levy classes? The average motorist levy increases by 63% while earners and employers respectively get 55% and 52% increases due to system-wide funding adjustments. This loads a disproportionate share of system-wide costs onto motorists. Why have you done it this way?

Secondly, within motor vehicle account, what kind of road-user mix are you trying to promote? The levies you propose for 2010/11:

• are based on a victim-pays mindset for levy-paying victims of road accidents;
• load a disproportionate share of the costs of non-levy-paying victims onto motorcycles; and
• over-charge light trucks and under-charge heavy trucks.

These levies affect the road use choices of New Zealanders. They convey a clear but unfortunate message: encase yourself in as much steel as possible.

When a car hits a motorcycle, it is the biker who tends to need more expensive care. But if the car wasn’t there, it wouldn’t have happened. There are two sides to this event, both equally necessary. So why don’t we charge the motorcycle and car accounts half each of all these costs? Why does the most mashed up victim have to pay the most, when it could never have happened without the other guy? Doesn’t this also weaken the incentive for car drivers to take efficient amounts of care regarding motorcycles?

Also, do you have any evidence to support your apparent view that motorcycles are similar to cars and trucks in their same propensity to inflict costly collision damage on cyclists and pedestrians? By simply counting registered (i.e. levy-paying) vehicles and allocating each vehicle the same share of (un-levied) cyclist & pedestrian costs you further ignore the principle that collision costs be attributable equally to the (levied) participants. Its bad enough that you mis-allocate motorcyclists own costs, adding an excessive share of the un-levied classes, who motorcyclists rarely harm badly, is particularly galling.

Provided you account for collisions properly, I really don’t mind if you differentiate levies between motorcycle types by adding extra classes of levies. But since you have proven able and willing to do this for/to motorcycles, why can’t you also do it for trucks? This quote is from the actuarial advice:

The exercise to calculate risk relativities revealed a large difference in risk rates between trucks and smaller commercial vehicles (vans and utility vehicles (utes)), with trucks showing a higher level of risk. This is driven by higher average claim costs. Under the current class structure, trucks, goods vans and utility vehicles (utes) are not distinguished from each other; these vehicles belong to either ACC Class 5 or 9, depending only on their motive power (petrol or non-petrol). If there is a difference in risk between trucks and vans/utes, these vehicles cannot be charged different rates under the current class structure, due to system constraints.

Why not, since you could do it for motorcycles? Basic physics suggests that these must be about the most damaging of all vehicle types to collide with, so this class should be a high priority. This omission distorts efficient choices of vehicle size, unduly tilting the market in favour of larger vehicles.

For completeness, and I’m sure you’re sick of answering this one, I must also the one that some motorcyclist’s 10 year old kid is reputed to have figured out. Can you please explain your public assertions about cars subsidising motorcycles? You said that each car on average subsidises motorcyclists by $77 per year. Thats about $200m of subsidy. So the cost of caring for motorcylists would need to be the $13m of levies from motorcyclists plus the $200m subsidy = $213m. But actually your total bill for motorcylists costs was $62m. Was this just a mistake?

Thanks for listening. And by the way, I really do want answers to my questions.

Yours sincerely

John Small

This shows that the vast majority of the increases are to cover income losses and pre-1999 catchup. Scheme cost increases are pretty modest, as is the cost of claims. It also suggests that motorists have copped significantly bigger increases for these scheme-wide factors than the other groups.

Kind friends pointed me to Motor Vehicle Crashes in New Zealand 2008, a statistical report prepared by the Ministry of Transport. While fairly complete, it is not detailed enough for what I wanted. The main ommission is information on accidents involving two or more road users (eg car:motorcycle). Lots of accidents must be like this, even though a few of us are prone to just canning out without involving other road users!

Information on these events is needed to figure out the pattern of cross-subsidies between road users. When a drunk car driver cleans out a biker or pedestrian, the ACC costs should be allocated to cars, not to motorcycles or pedestrians.The MoT data is not detailed enough to get into these issues, but I’m sure ACC will know because they’ve analysed cross-subsidies between road users. Maybe they’ll get back to me today.

In the meantime, here are a few summary graphs. The MoT crash data include numbers of crashes, contributing factors, injuries and deaths by road-user type. Here is the basic breakdown for 2008 of road accident victims:

Remember this is about who gets hurt/killed, not how it happened. Most of the victims are in steel-clad vehicles. Bikers and others (cyclists and pedestrians) cop it about equally. My guess is that, to the extent there are other parties involved, “other” road users are mostly being harmed by cars & trucks, not motorcyclists.

Here are the most important factors suspected of contributing to injuries, expressed as the total number of injuries for each user type (the data do not include information on pedestrians, so I’m only reporting info for bikers and a combination category of cars/trucks etc.

Its interesting that most of these factors are more likely to contribute to car/truck injuries than motorcycle injuries. The things that get bikers more often are loss of control and inexperience (which is probably correlated with loss of control – there can be more than one contributing factor). Also interesting that booze is much more likely to be a factor in car/truck injuries than for bikers.

Now here is the same information for things that contribute to deaths:

Notice that the horizontal scale goes a lot further for the death data. They show that booze was a suspected factor in 90% of motorcyclist deaths, a huge effect. But it doesn’t say who was drinking. Could be a Darwin effect (drunk biker kills themselves) or a collision with a car/truck driven by a drunk. Loss of control and excessive speed are also big factors, with inexperience also being important.

The advice is for the whole motor vehicle account within ACC. It starts off arguing that ACC needs $417 per vehicle to cover its costs. Then it splits this revenue requirement between licensing and petrol sales, both of which are taxed. An obvious way to have avoided the outrage would have been to get a higher share of revenue from petrol tax, but the split between license and petrol revenue is not discussed in the advice.

Then we get to the interesting stuff – how this $417 per vehicle will be shared across vehicle types. ACC say they have examined “the cost of injuries for motorcycles”  and while those costs are not reported, that analysis leads to the reclassification: mopeds come into the pool and there are 3 engine size bands.

Next comes relativities between classes. ACC has

“…reviewed its class relativities this year. This is intended to understand and reduce the level of cross-subsidisation between vehicle classes (p24/5)

Sounds good, but unfortunately they can’t do this directly.

“…ACC does not collect information on the precise vehicle class associated with each Motor Vehicle Account claims. (p25)

“…The CAS information only allows the number of claims in each class to be identified, rather than the claim cost. (p25)

This casts doubt on the accuracy of the earlier analysis of motorcycle injuries, and of the proposed split between bikes and other vehicles. The advice ploughs on into a statistical procedure they used to get some insight into actual claim costs per class, but none of this could be replicated using the information in the report, so a fair bit of doubt remains.

After “relativities” pop out of the statistical sausage machine, some weighting is done.

“…Relativities have been estimated for each accident year since the 2003/04 year, and these relativities are then weighted by vehicle exposure to obtain an overall relativity. A 50% weighting has been selected for the 2008/09 accident year, to reflect the higher level of uncertainty contained in these estimates.  (p27)

Its nice to see that they are acknowledging the uncertainty, but this weighting process is even more obscure than the statistical modelling. A table is presented showing percentages for classes 4 and 5 relative to cars (class 2). Then the advice comments about trends in classes 5/9, even though classes above 5 are not cited in the table. Its all a bit rough and ready.

Then we find out that trucks are getting a sweet deal:

“…The exercise to calculate risk relativities revealed a large difference in risk rates between trucks and smaller commercial vehicles (vans and utility vehicles (utes)), with trucks showing a higher level of risk. This is driven by higher average claim costs. Under the current class structure, trucks, goods vans and utility vehicles (utes) are not distinguished from each other; these vehicles belong to either ACC Class 5 or 9, depending only on their motive power (petrol or non-petrol). If there is a difference in risk between trucks and vans/utes, these vehicles cannot be charged different rates under the current class structure, due to system constraints. (p28)

Great eh? They’re able to create new classes for the purpose of kicking motorcyclists, but “system constraints” stop them doing this for trucks.

Finally, we get told that all of the above are just estimates of the true relativities between ACC vehicle classes, but are not necessarily the rates to be adopted by ACC when setting levies. In particular, ACC needs to ensure that its levies:

• Are appropriate when other factors are taken into consideration, and are reasonable
• Do not fluctuate unduly
• Reflect the appropriate level of cross-subsidisation between vehicle classes. There may be other external reasons to justify a certain level of cross-subsidisation (p28)

Apparently, if the calculated rates were used, motorcyclists would face levy rates far in excess of the current levy rates. So they’ve been nice to bikers, and cut them back dramatically, sticking with the historic relativity which has bikers paying 1.5 times what cars pay on average, but using their motorcycle engine size analysis to load most of that onto big bikes. Another interpretation of this decision could be that they are not very confident of their modelling.

Finally, we get this table that I found utterly impenetrable. It supposedly refers to petrol levies by vehicle class. “The licence levy rates for each class of petrol vehicles under this scenario is calculated as:

I don’t get this at all. We all pay 9.9c/l, not different prices depending on our vehicle.

Anyway, lets try to wrap this up with a few conclusions.

1. The ACC has obviously been focusing on motorcycle injuries, and feels that it has made some progress in understanding their pattern. The report is not at all convincing however. It could and should have been released with much more information about claim costs by class of vehicle.

2. There is a bias against motorcycles embedded deeply within this whole system. Like pedestrians, we tend to hurt more than car users after a road accident, and maybe we cost more to fix. But unlike pedestrians, we get taxed. So the more we hurt, the more we pay. Its especially galling to see that trucks get a sweet deal since they have the greatest potential to hurt the rest of us.

3. The financial pain is partly a result of the way ACC has chosen to split its revenue requirement between license fees and petrol levies. ACC should justify its split but has made no attempt at doing so, even though they acknowledge that exposure is related to use of roads rather than ownership of vehicles.

That’s enough for now – I’m off for a ride.